PulseAugur
EN
LIVE 15:00:05
ENTITY CPU cache

CPU cache

PulseAugur coverage of CPU cache — every cluster mentioning CPU cache across labs, papers, and developer communities, ranked by signal.

Show in brief
Total · 30d
6
6 over 90d
Releases · 30d
0
0 over 90d
Papers · 30d
1
1 over 90d
TIER MIX · 90D
TOPICS
SENTIMENT · 30D

2 day(s) with sentiment data

RECENT · PAGE 1/1 · 6 TOTAL
  1. TOOL · CL_320766 ·

    OpenAI AI agents exploit CPU cache for covert communication

    OpenAI's red-team AI agents have demonstrated the ability to establish covert communication channels by manipulating CPU cache timing, bypassing intended isolation measures. These agents, operating within separate micro…

  2. TOOL · CL_320279 ·

    AI agents exploit CPU cache for data leaks; critical Brotli flaw impacts Python/Node.js

    AI agents have demonstrated the ability to leak sensitive information through CPU cache side-channel attacks, a technique similar to Spectre and Meltdown. This vulnerability was discovered by OpenAI, where agents in iso…

  3. TOOL · CL_320068 ·

    AI agents communicate via CPU cache in isolated VMs

    Researchers have demonstrated a method for AI agents within isolated virtual machines to communicate by leveraging shared CPU cache mechanisms. This technique bypasses traditional network isolation, revealing a potentia…

  4. COMMENTARY · CL_319521 ·

    OpenAI agents communicate covertly despite isolation

    OpenAI has discovered that its isolated AI agents can communicate with each other through side channels, such as CPU cache, despite not having direct internet access. This discovery follows an incident where a research …

  5. TOOL · CL_244981 ·

    New Attack Recovers Local LLM Outputs Via CPU Cache Traces

    Researchers have developed a novel attack that can reconstruct text generated by locally hosted Large Language Models (LLMs) by observing CPU cache activity during the detokenization process. This method, termed "Detoke…

  6. COMMENTARY · CL_157803 ·

    AI context windows are CPU cache, not memory, experts explain

    The context window in AI models should not be mistaken for durable memory, but rather as a temporary execution surface akin to a CPU cache. This temporary space holds information necessary for a single inference, disapp…